Test of Significance

Test of Significance of a Trend
Name
Institution
Test of Significance of a Trend
A significance test may be described as one that challenges an established hypothesis in an attempt to prove whether the alternative hypothesis gives a pre-established level of significance. The significance tests seek to invalidate the perception of chance and discard a null hypothesis through following a given set of observed principles. A trend, on the other, is the correlation or association between concentration and spatial location or time. A trend can also be any population characteristic that changes in a given predictable way with a given variable (Şen, 2017). Trends take different forms and may be periodic, decreasing, or increasing. Noticing and gauging spatial and temporal trends are critical to monitoring programs and environmental studies.
Trend tests can be deployed as an alternative to control charts or prediction limits in the detection of monitors. The trend evaluation is then used to establish if it is sensible to assume concentrations are momentarily stationary and to model or detect declining trends in an attempt to support studies of natural attenuation. Linear regressions may also be used for testing of temporal trends. This requires the regression of ordinary least squares when fitting the best line. When the gradient of the linear correlation is not zero, a linear trend is obtained, with a positive gradient indicating an upsurge in trend and a negative slope showing a reduction in trend (Şen, 2017).

However much the model used in forecasting concentrations at future times with the assumption that a similar linear relationship will be obtained, there is a high likelihood that the assumption may not be valid with monitoring distant future events. For a decreasing trend, while testing significance, the line or regression should be inferred to approximate the time a criterion is likely to be met. However, when the regression line is assumed to be valid for future monitoring, the approach may not give conservative estimates as it does not consider regression uncertainty fit resulted to data variability within calculated line of regression.
Reference
Şen, Z. (2017). Statistical Trend Tests. In Innovative Trend Methodologies in Science and Engineering (pp. 67-132). Springer, Cham.